# FFLAME: a fragment-to-framework learning approach for MOF potentials

## TL;DR

FFLAME is a new method for predicting properties of metal-organic frameworks (MOFs) using machine learning by focusing on their building blocks, making predictions more accurate and efficient.

## Contribution

FFLAME introduces a fragment-based approach to train transferable machine learning potentials for MOFs, improving generalizability and reducing data requirements.

## Key findings

- Fragment-informed training improves model generalizability in data-scarce scenarios.
- FFLAME achieves near-target accuracy on unseen MOFs with minimal additional training.
- The approach accelerates convergence during fine-tuning and reduces the need for full-framework data.

## Abstract

Metal–organic frameworks (MOFs) exhibit immense structural diversity and hold promise for applications ranging from gas storage and separation to energy storage and conversion. However, structural flexibility makes accurate and scalable property prediction difficult. While machine learning potentials (MLPs) offer a compelling balance between accuracy and efficiency, most existing models are system-specific and lack transferability across different MOFs. In this work, we introduce FFLAME – Fragment-to-Framework Learning Approach for MOF Potentials, a fragment-centric strategy for training transferable MLPs. By decomposing MOFs into their constituent metal clusters and organic linkers, FFLAME enables efficient reuse of chemical environments and significantly reduces the need for full-framework training data. We demonstrate that fragment-informed training improves model generalizability, particularly in data-scarce regimes, and accelerates convergence during fine-tuning. FFLAME achieves near-target accuracy on unseen MOFs with minimal additional training. These results establish a robust and data-efficient pathway toward general-purpose MLPs for the simulation of diverse framework materials.

FFLAME, a fragment-centric strategy for training transferable MOF machine learning potentials, learns from building blocks, lowers data needs, and achieves near-target accuracy with minimal fine-tuning even for unseen MOFs.